Can we just link to /guides/ instead? Seems like we have two mostly-identical copies of the same page now?

(Code seems fine, I'm just not sure what the distinction/plan is on Config → Welcome vs Guides -- seems like Config → Welcome is completely obsolete now?)

I can un-prototype guides and just link to it from Config, I just hadn't thought through all the cases for that - like /welcome/ is the default for guides, but it's written for Phabricator, not Phacility. I think I can make guide modules on.... CORE? and have them pull in? Not sure where Phacility modules would live and get pulled in.

That seems reasonable to me.

Phacility Modules should either go in Services (rSERVICES) ("How to use your new Phabricator instance") or possibly in the far future in Instances (rSAAS) ("How to use the Phacility administration console").

There were also 2 TODOs here, I could use a short reply on. Mostly - "Go check out Applications" seems exceedingly vague, and no real means of verifying success, but still useful to point to the page.

On actual content:

I don't think we should instruct users to uninstall applications. I think this is an unusual workflow (for example, when you install OSX or Windows the first thing you see isn't a prompt to go delete stuff), it feels weird, and it's bad for us and for users. I don't think having applications installed should be hurting anything. The only case where I'd expect installs to actually remove an application is in fairly rare cases to actively prevent users from using it (e.g., Legal is uncomfortable with chat being available in Phabricator or whatever because of international subpoena laws).

In most cases, a large number of Applications should be able to exist without hurting anything. When you install an OS, you get tons of apps and that's fine. Your phone comes with tons of apps too. In the future, installs should be able to add a hundred third-party applications without hurting the experience for users. Although you might go organize your phone a bit at some point, I don't think anyone goes and deletes Calculator.exe after installing Windows because they use Google Calculator instead. If there are cases where having lots of applications is harmful, we should try to fix those problems without requiring users to uninstall applications (D16459 is a small change in this direction).

One such problem case is that the large number of applications can possibly be overwhelming to new users, but I don't think that's inherently a problem of having a lot of applications, and that's what this workflow should fix. This workflow should be able to guide users to accomplish things without requiring them to uninstall applications.

We could alternately test to see if users have configured any application (e.g., changed policies) but I don't think first-time administrators are likely to be able to understand how/why to do this, and especially not as a first step.

We could do something vague ( "Explore Applications") but currently have no way to verify success.

And then, per T11132#191013, I think:

• We should not put "Invite Collaborators" above content creation. I think first-time administrators are likely to want to create content before they show anyone, and that new invites will have a better experience if they reach an install with content. It's way better to get an invite from a first-time administrator who has already convinced themselves that they like Phabricator, added a repository or two, created a relevant dashboard, and maybe personalized the install, and wants to show that stuff off than an invite from a first-time administrator who isn't familiar with the software yet to join an empty instance that they aren't sold on. "Invite Collaborators" should be near the end: "show off all your work".
• We should not prompt first-time administrators to create projects. Particularly as written, I think this step will teach the wrong lessons and reinforce an incorrect model for users coming from JIRA.

I'd guess our completion rate on "Build a Dashboard" as written would also be close to 0%. To complete the quest you currently need to create a dashboard and install it on the home page, which we don't tell you explicitly and don't give you any guidance on.

The shortest path I see here without doing engineering NUX work is maybe to turn "Applications" and "Dashboards" into not-auto-detected guidance paragraphs that are just static text with a link to the thing and not completable? Not great, but better than nothing, which is basically what we have today.

The People check is still fine on Phacility in terms of looking for completion (when users accept invitations, accounts get generated), the link to /people/invite/ is the only issue. To fix that I think we have to do more modularization so Phacility can turn off the upstream "invite" quest and put a quest of its own in its place.